The Duplicated α7 Subunits Assemble and Form Functional Nicotinic Receptors with the Full-length α7*

Background: Schizophrenia is linked to the α7 nicotinic acetylcholine receptor and to duplicated genes encoding truncated dupα7 and dupΔα7 subunits. Results: Fluorescently labeled duplicated subunits display FRET with α7. Electrophysiology shows that duplicated subunits and α7 form heteropentamers with altered responses to choline and varenicline. Conclusion: Duplicated subunits co-assemble with α7, forming functional receptors. Significance: Such heteropentameric receptors could play a role in schizophrenia. The α7 nicotinic acetylcholine receptor gene (CHRNA7) is linked to schizophrenia. A partial duplication of CHRNA7 (CHRFAM7A) is found in humans on 15q13–14. Exon 6 of CHRFAM7A harbors a 2-bp deletion polymorphism, CHRFAM7AΔ2bp, which is also associated with schizophrenia. To understand the effects of the duplicated subunits on α7 receptors, we fused α7, dupα7, and dupΔα7 subunits with various fluorescent proteins. The duplicated subunits co-localized with full-length α7 subunits in mouse neuroblastoma cells (Neuro2a) as well as rat hippocampal neurons. We investigated the interaction between the duplicated subunits and full-length α7 by measuring Förster resonance energy transfer using donor recovery after photobleaching and fluorescence lifetime imaging microscopy. The results revealed that the duplicated proteins co-assemble with α7. In electrophysiological studies, Leu at the 9′-position in the M2 membrane-spanning segment was replaced with Cys in dupα7 or dupΔα7, and constructs were co-transfected with full-length α7 in Neuro2a cells. Exposure to ethylammonium methanethiosulfonate inhibited acetylcholine-induced currents, showing that the assembled functional nicotinic acetylcholine receptors (nAChRs) included the duplicated subunit. Incorporation of dupα7 and dupΔα7 subunits modestly changes the sensitivity of receptors to choline and varenicline. Thus, the duplicated proteins are assembled and transported to the cell membrane together with full-length α7 subunits and alter the function of the nAChRs. The characterization of dupα7 and dupΔα7 as well as their influence on α7 nAChRs may help explain the pathophysiology of schizophrenia and may suggest therapeutic strategies.

Schizophrenia is a complex neuropsychiatric disorder. Its heterogeneity, incomplete penetrance, and environmental factors provide challenges for identifying genetic defects. On-line Mendelian Inheritance in Man now lists more than 200 candi-date genes for schizophrenia susceptibility. Many genes have been associated with schizophrenia, yet there is no definitive genetic model. The etiology may involve a set of genes predisposing to the illness (1,2).
One of the replicated candidate genes for schizophrenia is the ␣7 nicotinic acetylcholine receptor (nAChR) 2 subunit gene, CHRNA7. Both pharmacological and genetic studies support a potential link between the ␣7 nAChR and the pathophysiology of schizophrenia. The expression of the ␣7 receptor, as measured by ␣-bungarotoxin binding, is decreased in postmortem hippocampus, cortex, and thalamus of schizophrenic subjects (3)(4)(5)(6). Moreover, CHRNA7 localized at 15q14 is linked to the P50 auditory sensory processing deficit in schizophrenia by genetic and neurobiological studies (7,8). Polymorphisms in the 5Ј upstream regulatory region and intron 2 of CHRNA7 are associated with both schizophrenia and the sensory processing deficit (9,10).
The human CHRNA7 gene is partially duplicated. Exons 5-10 of CHRNA7 and additional DNA (ϳ200 kbp) were duplicated upstream, interrupting a second partial duplication of four exons from the gene ULK4 on chromosome 3 (11,12). This duplication appears to be a relatively recent event unique to humans (13). The partially duplicated chimeric gene, CHRFAM7A, is located at 15q14 1.6 Mb centromeric to the full-length gene, and it is almost always in the opposite orientation to CHRNA7. In this duplication, exons 5-10, intervening introns, and the 3Ј-untranslated region of CHRNA7 are conserved. The gene product, dup␣7, lacks the signal peptide and part of the binding site but contains all of the ␣7 membranespanning regions. Mutation screening in human DNA and mRNA identified a 2-bp deletion polymorphism in exon 6 of CHRFAM7A (14). The partial duplication with the 2-bp deletion, CHRFAM7A⌬2bp, is almost always in the same orienta-tion as CHRNA7 (15). The 2-bp deletion in exon 6 causes a frameshift in translation, resulting in a smaller gene product, dup⌬␣7, which lacks nearly the entire ligand-binding site compared with dup␣7. CHRFAM7A⌬2bp is significantly associated with schizophrenia (16,17), suggesting that the duplicated gene might contribute to cognitive impairment.
The gene products of CHRFAM7A, both mRNA and protein, are found in brain and also in the periphery (11). When expressed alone in cell lines or Xenopus oocytes heterologously, the protein product of CHRFAM7A, dup␣7, is detected at low levels, but no functional receptor is observed (18). When either dup␣7 or dup⌬␣7, the protein product of CHRFAM7A⌬2bp, is co-expressed with ␣7 in Xenopus oocytes, it reduces the amplitude of the acetylcholine (ACh)-evoked currents, and two previous studies show that the reduction increases with the dose of dup␣7 (19,20). The pharmacology of the expressed receptors shows differences from that of the ␣7 heteromer (20). Immunological studies in GH4C1 cells show that dup␣7 reaches the plasma membrane (20). The authors of these previous studies suggest that CHRFAM7A acts as a dominant negative modulator of CHRNA7 function in humans and thus may be critical for receptor regulation (19,20).
This study further characterizes the role of CHRFAM7A by investigating how dup␣7 and dup⌬␣7 exert their effects on the full-length ␣7 nAChR. We have studied the interaction between fluorescently labeled duplicated and full-length subunits in mouse neuroblastoma cells (Neuro2a) using imaging techniques. We concentrated on the possible stoichiometry of the assembled complexes. We have also performed electrophysiology recordings to examine the receptor properties. We confirm that the duplicated subunits are assembled and transported to the cell membrane together with full-length ␣7 and thus alter the function of the nAChRs. Our results contribute to the understanding of the pathophysiology of schizophrenia and suggest therapeutic strategies.

EXPERIMENTAL PROCEDURES
Reagents-Unless otherwise indicated, all chemicals for culture were purchased from Invitrogen, and all chemicals for electrophysiology were purchased from Sigma.
Plasmid Constructs-Human ␣7, dup␣7, and dup⌬␣7 cDNA clones in pcDNA 3.1 were described previously (19). The fluorescent protein (FP) cassette, containing an Ala-Gly-Ala linker flanking the FP open reading frame on both sides, was fused into the M3-M4 loop of the subunits (21). The fusion site was chosen as described previously (22). Mutations were introduced using the QuikChange kit from Stratagene (La Jolla, CA) according to the manufacturer's instructions.
Cell Culture and Transfections-Mouse Neuro2a cells obtained from American Type Culture Collection (ATCC) were maintained in 45% Dulbecco's modified Eagle's medium (DMEM), 45% Opti-MEM, 10% fetal bovine serum (FBS) and supplemented with penicillin and streptomycin. Neuro2a cells were transfected with BioT (Bioland Scientific LLC, Paramount, CA) according to the manufacturer's instructions. Plasmid concentrations used for transfection were as follows: for imaging, 1 g of each fluorescently labeled nAChR subunit; for competition experiments, 500 ng of each fluorescently labeled subunit ϩ 1 g of untagged subunit; for recording, 1 g of each fluorescently labeled nAChR subunit ϩ 200 ng of RIC-3. Rat hippocampal neurons were dissociated from day 18 rat embryos and plated on 35-mm glass-bottomed imaging dishes (MatTek, Ashland, MA) as described previously (23). Neurons were transfected after 7 days in culture using Lipofectamine 2000 in conjunction with Nupherin (Biomol Research Laboratories, Plymouth Meeting, PA) according to the manufacturer's instructions (24). 1 g of each fluorescently labeled subunit was used.
Confocal Imaging-Neuro2a cells were plated at 90,000 cells per dish and imaged live 48 h after transfection. Before imaging, cell culture medium was replaced with phenol red-free CO 2independent Leibovitz L-15 medium. Cells were imaged with a Nikon (Nikon Instruments, Melville, NY) C1 laser-scanning confocal microscope system equipped with a 63 ϫ 1.4 numerical aperture VC Plan Apo oil objective. Pinhole diameter was 30 -60 m, and cells were imaged at 12-bit intensity resolution over 512 ϫ 512 pixels at a pixel dwell time of 6.12 s. Cellular mGFP and mCherry fluorescent signals were acquired by simultaneous excitation with 488-nm (for mGFP) and 561-nm (for mCherry) lasers. Emission spectra were acquired in 5-nm bins between 500 and 660 nm, and the signal of each expressed fluorophore was linearly unmixed from the raw spectral image using reference spectra from control cells expressing only mGFP or only mCherry fusion constructs. For each pixel of a spectral image, intensity of mGFP and mCherry was determined from fluorescence intensity values at the peak emission wavelength derived from the reference spectra.
Förster Resonance Energy Transfer (FRET) by Donor Recovery after Acceptor Photobleaching (DRAP)-To examine FRET between nAChR subunits, the acceptor photobleaching method (21) was used with a modified DRAP macro built into the EZC1 imaging software (Nikon). The fluorescence intensity of mGFP and mCherry was recorded simultaneously using a 488-and 561-nm laser, respectively, during a six-step incremental photodestruction of mCherry with a 561-nm laser and normalized to the prebleach time point. I D value was extrapolated from a scatterplot of the fractional increase of mGFP versus the fractional decrease of mCherry. FRET efficiency was calculated as E ϭ 1 Ϫ (I DA /I D ), where I DA represents the normalized fluorescence intensity of mGFP (100%) in the presence of both donor (mGFP) and acceptor (mCherry), and I D represents the normalized fluorescence intensity of mGFP in the presence of donor only (complete photobleaching of mCherry). Data were averaged from 20 to 30 cells per condition and reported as mean Ϯ S.E.
FRET measurements may include contributions from pairs of fluorophores that are nearby but not in a macromolecular complex. In the biophysical literature, this has also been termed stochastic FRET, bystander FRET, background FRET, and proximity FRET. King et al. (25) calculated and measured that proximity FRET became negligible for tetramers at membrane densities less than ϳ10 Ϫ3 receptors/nm 2 or 10 3 receptors/m 2 , and similar estimates would apply to pentamers. The closely packed acetylcholine receptors at the nerve-muscle synapse have densities of 10 4 /m 2 . For this study, densities on the plasma membrane were probably Ͻ10 2 /m 2 , judging by the ␣7/dup␣7 and ␣7/dup⌬␣7 Heteropentameric nAChRs small size of the currents we observed, by the fact that receptors were not evident in membrane profiles, and by the fact that we have rarely observed signals from single ␣7 nAChRs in membrane patches studied by zero-mode waveguides (26); measurements from Simonson et al. (27) yielded a similar conclusion. A more substantial, but less measurable, source of stochastic FRET would arise from fluorophores in the endoplasmic reticulum membrane. Total internal reflection fluorescent measurements of endoplasmic reticulum-resident heteromeric nAChRs (28) have led us to the assumption that these densities are Ͻ10 3 /m 2 ; and the much lower signals observed in this study led us to an estimate of Ͻ10 2 /m 2 in the endoplasmic reticulum. Thus, we conclude that proximity FRET has played little or no role in our measurements. An abundance of caution leads us to assume that in DRAP mode, FRET efficiencies of Ͻ10% may include a contribution from proximity FRET.
FRET by Fluorescence Lifetime Imaging Microscopy (FLIM)-We have monitored the fluorescence lifetime by time-correlated single-photon counting. YFP (donor)-and mCherry (acceptor)-tagged nAChRs were co-expressed in Neuro2a cells. We excited YFP with a picosecond pulsed laser (514 nm) and monitored the fluorescence emission using a time-correlated single-photon counting module (PicoQuant, West Springfield, MA). Data were collected at 530 Ϯ 20 nm through a bandpass filter. Laser power was adjusted to give average photon-counting rates of the order 10 4 to 10 5 photons/s to avoid pulse pile up. The decay of the YFP fluorescent intensity was fitted into either a single exponential (when YFP is present alone) or a double exponential (when YFP and mCherry are present) with Sym-PhoTime software (PicoQuant) using Equation 1, where 1 is the fluorescence lifetime of a YFP molecule when it is present alone, and 2 is the fluorescence lifetime of a YFP molecule when it interacts with its acceptor mCherry. We calculated the "binding fraction" and FRET efficiency using Equations 2 and 3, where P FRET is the binding fraction, the fraction of YFP molecules that interact with mCherry, and E FRET is the FRET efficiency.
Whole-cell Patch Clamp Recordings-Neuro2a cells were plated on 12-mm glass coverslips within 35-mm plastic bottom cell culture dishes at the density of 50,000 cells per dish. 48 -72 h after transfection, glass coverslips were transferred to the recording chamber on the microscope stage. Expressing cells were identified and visualized as follows. Green (mGFP) and red (mCherry) fluorescence was visualized with an upright microscope (BX50WI, Olympus) and UV illumination with appropriate excitation filters. Cells were voltage-clamped at a holding potential of Ϫ65 mV. Agonists were applied using a focal drug application system to minimize desensitization (29).
For concentration-response studies, ACh was applied from the lowest to highest concentration at 3-min intervals to get the dose-response curve. Whole-cell patch clamp recordings were performed with a MultiClamp 700B amplifier, a 1322 analogto-digital converter, and pCLAMP 9.2 software (all from Axon Instruments, Molecular Devices). Data were sampled at 10 kHz and filtered at 2 kHz.
The intrapipette solution contained (in mM) the following: 135 potassium gluconate, 5 KCl, 5 EGTA, 0.5 CaCl 2 , 10 HEPES, 2 Mg-ATP, and 0.1 GTP; the pH was adjusted to 7.2 with Tris base and the osmolarity to 300 mOsM with sucrose. The extracellular solution contained (in mM) the following: 140 NaCl, 5 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 HEPES, and 10 glucose; pH was adjusted to 7.3 with Tris base. The Nernst potential for Cl Ϫ in the intrapipette solution is Ϫ82.9 mV. The bath was continuously perfused with extracellular solution at room temperature. The patch electrodes had resistances of 5-8 megohms. The junction potential between the patch pipette and the bath solutions was nulled just before forming a seal. Series resistance was monitored and compensated by 70 -80% throughout the recordings. The data were ignored if the series resistance (15-30 megohms) changed by Ͼ20% during the recording session.
For the substituted cysteine accessibility method (SCAM), each cell received one initial baseline application of ACh (300 M), one control ACh application, an application of ethylammonium methanethiosulfonate (MTSEA, 2 mM, Toronto Research Chemicals Inc., North York, ON, Canada), and then a follow-up experimental application of ACh. The peak amplitude of control responses and experimental responses were calculated relative to baseline ACh responses to normalize the data.

Co-localization of Fluorescently Labeled dup␣7/dup⌬␣7
Subunits and the Full-length ␣7-Based on previous work (21,22), we fused fluorescent protein to the M3-M4 loop of fulllength human ␣7 nAChRs as well as the dup␣7 and dup⌬␣7 subunits (Fig. 1A). Fluorescently labeled subunits are co-expressed in Neuro2a cells and imaged alive using a confocal microscope (Fig. 1B). We noted that the heterologous expression of duplicated subunits was low. The chimeric gene, CHRFAM7A, is transcribed efficiently, but translation is poor (19). The intensity of the FP attached to these subunits was ϳ5% of the intensity of the FP attached to the full-length subunits. When co-expressed with fluorescent ␣7, the duplicated subunits were localized primarily in the endoplasmic reticulum, similar to the localization of the full-length receptors. We did not systematically study localization at the plasma membrane. The merged images of the GFP and mCherry signals indicate that both dup␣7 and dup⌬␣7 subunits co-localize with the full-length ␣7 receptor very well. Pearson correlation coefficients were as follows: ␣7GFP with dup␣7mCherry, 0.96 (n ϭ 26); ␣7GFP with dup⌬␣7mCherry, 0.95 (n ϭ 25); dup␣7GFP with ␣7mCherry, 0.96 (n ϭ 27); and dup⌬␣7GFP with ␣7mCherry, 0.97 (n ϭ 26). We then co-expressed the duplicated and the full-length subunits in primary rat hippocampal neurons. Because of the low intensity of the labeled duplicated subunits, they were not detected in the distal sites such as axons and dendrites. In the cell soma, where we were able to detect the duplicated subunits, both dup␣7 and dup⌬␣7 were co-localized with the full-length ␣7 (Fig. 1C). Pearson correlation coefficients were as follows: ␣7GFP with dup␣7mCherry, 0.85 (n ϭ 11); ␣7GFP with dup⌬␣7mCherry, 0.86 (n ϭ 11); dup␣7GFP with ␣7mCherry, 0.93 (n ϭ 10); and dup⌬␣7GFP with ␣7mCherry, 0.91 (n ϭ 12).
Interaction between dup␣7/dup⌬␣7 Subunits and the Fulllength ␣7, DRAP Assays-The fact that the duplicated and the full-length subunits are co-localized both in Neuro2a cells and in the cell body of neurons suggests that they may be assembled into oligomers. The receptor assembly of nicotinic subunits is often measured by immunoprecipitation (30). This approach has limited applicability, however, for dup␣7 and dup⌬␣7, due both to the extremely low protein levels and the lack of immune reagents selective for the duplicated subunits.
To directly determine whether the duplicated subunits interact with the full-length ␣7, we measured FRET between fluorescently tagged receptor subunits. FRET occurs only when donors and acceptors are within ϳ100 Å. When the fluorescent proteins are fused into the M3-M4 loop of nicotinic receptor subunits, subunits within pentamers undergo FRET. We have previously developed experimental approaches, well supported by theory, to analyze the assembly and subunit stoichiometry of such subunits within Cys loop receptors (23,28,(31)(32)(33). For the present experiments, important guidelines arise from the strong distance dependence of FRET. Thus, for DRAP experiments adjacent subunits have considerably higher FRET values than nonadjacent subunits, and a donor adjacent to two acceptors displays higher FRET than one adjacent to zero or one acceptor.
We first employed DRAP to measure FRET between subunits (34). An mGFP/mCherry FRET pair was used for good separation of the excitation and emission spectra. We incrementally bleached mCherry at 561 nm while monitoring fluorescence intensity of both mGFP and mCherry, at 488 and 561 nm, respectively. Changes in fluorescence intensity versus time were plotted for each cell, and FRET efficiency was calculated as described under "Experimental Procedures." DRAP data revealed several types of co-assembly as follows: ␣7 with ␣7; ␣7 with dup␣7; ␣7 with dup⌬␣7; dup␣7 with dup␣7; and dup⌬␣7 with dup⌬␣7 ( Fig. 2A). The measured FRET efficiencies were different between reciprocal pairs. FRET efficiency was greater when dup␣7, or dup⌬␣7, was the donor. This is likely due to the low expression of the duplicated gene; a dup␣7 subunit is more likely to lie adjacent to an ␣7 subunit in the heteromeric receptor.
To confirm the interaction between the duplicated subunits and the full-length subunit, we measured FRET between ␣7-mGFP and ␣7-mCherry when untagged competing subunits were co-expressed. Calculated FRET efficiency showed that co-expression of untagged subunits, either duplicated or full-length, significantly decreased the FRET efficiency of the ␣7-mGFP/␣7-mCherry pair (Fig. 2B), suggesting the specific interaction between dup␣7/dup⌬␣7 subunits and the fulllength ␣7. ␣7 is full-length human ␣7 nAChR; dup␣7 is partial duplication of human ␣7; dup⌬␣7 is partial duplication of human ␣7 with a 2-bp deletion. Black boxes represent transmembrane domains; gray box represents duplication of sequences from gene ULK4 on chromosome 3, and striped boxes represent fluorescent proteins fused into the M3-M4 loop of nAChRs. B, duplicated subunits and full-length ␣7 receptors are localized similarly in Neuro2a cells. Neuro2a cells were transfected with the indicated nAChR cDNAs and were imaged live with spectral confocal microscopy 48 h after transfection. Spectral images were acquired, and specific mGFP and mCherry signals were extracted with linear unmixing. Fluorophore abbreviations are as follows: G, mGFP; Ch, mCherry. Green (mGFP signal) and red (mCherry signal) pseudocolor was assigned, and yellow (Merge) indicates co-localized proteins. C, duplicated subunits and full-length ␣7 receptors are localized similarly in the cell soma in primary neurons. E18 rat hippocampal neurons were plated and cultured for 14 days followed by transfection with the indicated nAChR cDNAs. One day after transfection, cells were imaged live using confocal microscopy. Normalization was done to emphasize co-localization and differed for the full-length versus duplicated constructs. Without normalization, the merged signal is mostly green (when ␣7mGFP is expressed) or red (when ␣7mCherry is expressed) and provides little information about co-localization. All images in C are less bright than B because the overall protein expression level is lower in primary neurons (C) than in Neuro2a cells (B). Scale bar, 10 m. ␣7/dup␣7 and ␣7/dup⌬␣7 Heteropentameric nAChRs Potential Stoichiometry of ␣7dup␣7 and ␣7dup⌬␣7 Heteromers, FLIM Measurements-DRAP is not optimal for determining subunit stoichiometry in the present experiments because there are many more full-length ␣7 subunits than duplicated subunits when they are co-expressed (32). We also employed FLIM to measure FRET. The fluorescence decay profile of the donor, YFP molecules, was acquired from the entire cell with the use of a time-and space-correlated single-photon counting detector and was fitted to a double exponential time course (Fig. 3A). Neuro2a cells transfected with various nAChR cDNAs were monitored alive. The binding fraction, the fraction of donor that interacts with the acceptor, and FRET efficiency were calculated as described under "Experimental Procedures." In the context of analyses presented elsewhere, we believe that the binding fraction is best interpreted as the fraction of donors adjacent to one or more acceptors (23,28,31,32). These two metrics are relatively insensitive to fluorophore concentration and light-path length, conditions that are poorly controlled inside a cell (35,36). FLIM data confirmed several types of co-assembly as follows: ␣7 with ␣7; ␣7 with dup␣7; ␣7 with dup⌬␣7; dup␣7 with dup␣7; and dup⌬␣7 with dup⌬␣7 (Fig. 3B).
To investigate the stoichiometry of ␣7/dup␣7 and ␣7/dup⌬␣7 heteromers, we measured FRET between various pairs when untagged competing subunits were co-expressed. Both binding fraction and FRET efficiency of the ␣7-YFP/␣7-mCherry pair were decreased, as expected, by untagged full-length ␣7 (Fig. 3, C 1 and C 2 ). Interestingly, co-expression of untagged duplicated ␣7 decreased binding fraction but not FRET efficiency of the ␣7-YFP/␣7-mCherry pair, suggesting a limited incorporation of the duplicated subunits. In an (␣7-YFP) 1 (␣7-mCherry) 4 configuration, as the number of untagged full-length subunits increases, the donor is less likely to be adjacent to an acceptor, which results in a decrease in the binding fraction. In contrast, if only one acceptor can be replaced by the duplicated subunit, the donor is still adjacent to an acceptor; thus one expects only a modest change in the binding fraction, and this is observed experimentally (Fig. 3C 1 ). Furthermore, when untagged fulllength ␣7 was co-expressed for competition, both binding fraction and FRET efficiency of dup␣7-YFP/␣7-mCherry and dup⌬␣7-YFP/␣7-mCherry pairs were decreased significantly. The difference between the reciprocal conditions, dup␣7 competing for ␣7-YFP/␣7-mCherry pair and ␣7 competing for dup␣7-YFP/␣7-mCherry pair, also supports the limited incorporation of duplicated subunits; more than one acceptor can be replaced by untagged subunits when ␣7 is competing, decreasing FRET substantially.
Leucine at the 247-position of ␣7 nicotinic receptors resides ϳ9 residues downstream from the cytoplasmic beginning of the M2 pore-lining region (termed the 9Ј-position in most studies, because the M2 domain is rather well conserved among Cys loop receptors). The M2 domain of dup␣7 and dup⌬␣7 has the same sequence as ␣7 in this domain. We replaced the Leu-9Ј residue with cysteine (L9ЈC mutation) in dup␣7 and dup⌬␣7. We transfected these subunits individually with ␣7 subunits and measured receptor function with the whole-cell patch clamp technique. If dup␣7 and dup⌬␣7 subunits assemble with ␣7 subunits to form functional receptors, we expect the introduced cysteines in dup␣7 and dup⌬␣7 to be accessible for MTSEA and MTSEA to change receptor function. As the ␣7 subunit has a cysteine at the 116-position, to eliminate the possibility that MTS derivatives bind to the endogenous cysteine and distort the results, we replaced this cysteine with serine in all of the constructs. The aligning mutation in dup␣7 and dup⌬␣7 has different numbers but are termed C116S for consistency. Previous reports show that the C116S does not alter ␣7 receptor function (39).
Using the whole-cell patch clamp technique and a focal drug application system (29,40), we examined 300 M ACh-evoked inward current (I ACh ) at 3-min intervals (Fig. 5A 1 ). In cells satisfying the criterion of stable baseline responses (Ͻ15% variation in 10 min), we subsequently perfused the cell with standard extracellular solution for 3 min to wash out ACh, applied 2 mM MTSEA for 2 min, washed out MTSEA for 5 min, and puffed 300 M ACh to evoke I ACh (Fig. 5A 2 ). We compared I ACh before and after MTSEA application to test whether MTSEA changed receptor function. Changes in agonist-induced currents indi- . FLIM confirmed assembly of duplicated subunits with full-length ␣7. A, schematic of FRET occurring when a donor, the YFP (yellow-green cylinder) molecule attached to a nAChR subunit (gray cylinder), and an acceptor, the mCherry (red cylinder) molecule attached to a nAChR subunit, are in close proximity, usually Ͻ10 nm. The nonradiative transfer of excited state energy of the donor to the acceptor shortens the donor fluorescence lifetime and results in a faster decay in the fluorescence intensity of the donors. B 1 and B 2 , Neuro2a cells were transfected with the indicated nAChR subunits and were imaged alive using FLIM. Binding fraction (B 1 ) and FRET efficiency (B 2 ) were calculated as described under "Experimental Procedures." Several types of co-assembly were detected as follows: ␣7 with ␣7, ␣7 with dup␣7, ␣7 with dup⌬␣7, dup␣7 with dup␣7, and dup⌬␣7 with dup⌬␣7. C 1 and C 2 , additional expression of untagged competing subunits (black circles) decreased the binding fraction of the ␣7-YFP/␣7-mCherry pair. **, p Ͻ 0.01; *, p Ͻ 0.05 versus first condition. D 1 and D 2 , additional expression of untagged full-length ␣7 subunit decreased both binding fraction and FRET efficiency of the dup␣7-YFP/␣7-mCherry and dup⌬␣7-YFP/␣7-mCherry pairs. Error bars are Ϯ S.E., and n ϭ 20 -30 cells for each condition. **, p Ͻ 0.01; *, p Ͻ 0.05. ␣7/dup␣7 and ␣7/dup⌬␣7 Heteropentameric nAChRs cate that the receptors include subunits with the cysteine mutation.
We transfected Neuro2a cells with ␣7 DNA constructs containing the C116S mutation. As a control, we transfected ␣7-mCherry, ␣7-mCherry ϩ dup␣7-GFP, or ␣7-mCherry ϩ dup⌬␣7-GFP into Neuro2a cells and tested whether MTSEA changed receptor function. As expected, MTSEA did not change I ACh in the transfected cells (Fig. 5, B 1 , C 1 , and D 1 ), confirming that introducing the C116S mutation was sufficient to prevent MTSEA from affecting the receptor's function. We next transfected ␣7-mCherry with both C116S and L9ЈC mutations into Neuro2a cells. In this situation, all of the subunits in the functional receptors are accessible to MTSEA. Indeed, we observed that MTSEA inhibited I ACh by 54 Ϯ 7% (n ϭ 5, p ϭ 0.002) (Fig. 5, B 2 and E 1 ). These results indicated that the assay was valid to reveal whether cysteine-containing subunits exist in functional receptors.
Because 1 mM ACh evoked near-saturating currents, activating the great majority of ␣7 receptors (Fig. 6D), the size of 1 mM ACh-induced currents may correlate well with the number of functional receptors. We compared 1 mM ACh-induced currents in cells expressing ␣7 subunits alone, ␣7 ϩ dup␣7 subunits, and ␣7 ϩ dup⌬␣7 subunits 72 h after transfection. As illustrated in Fig. 6E, the peak amplitude of I ACh did not differ among groups of cells. These data suggest that dup␣7 and dup⌬␣7 did not alter the number of ␣7 receptors. We acknowledge that ascertainment bias could have been introduced by the process of identifying and studying the fluorescent cells.
We applied 100 M ACh and 100 M choline for relatively prolonged pulses lasting 2 s, whereas we measured voltageclamp currents (Fig. 8, A 1 and B 1 ). Such experiments represent the classical tests for desensitization kinetics of nAChRs. As shown in Fig. 8, A 2 and B 2 , we found that the presence of dup␣7 and dup⌬␣7 did not alter the desensitization kinetics of ␣7 receptors in the presence of ACh and choline. In all cases, desensitization was nearly complete during the pulse.

DISCUSSION
Exons 1-6 of CHRNA7 correspond to the receptor's extracellular N-terminal region, which contains the ligand-binding domain. Exons 7 and 8 correspond to the first three transmembrane regions, M1, M2 and M3, and M2 constitutes the ion channel of the receptor (44). Exons 9 and 10 encode its intracellular cytoplasmic loop, the fourth transmembrane region, M4, and the extracellular C terminus. CHRFAM7A, containing only exons 5-10 of CHRNA7, lacks the signal peptide as well as part of the extracellular N-terminal region, including one of the three N-glycosylation sites. This lack of a ligand-binding domain renders it unsurprising that dup␣7, when expressed alone in Xenopus oocytes or Neuro2a cells, fails to produce ACh-induced current. In addition, previous studies have demonstrated that the N-terminal extracellular domain of nAChR plays a leading role in mediating assembly and determines the specificity of the intersubunit recognition (45)(46)(47). This study confirms that in a heterologous expression system, a truncated nAChR subunit lacking both signal peptide and part of the N-terminal extracellular domain is 1) processed in a similar way as the full-length subunit; 2) assembled with the full-length subunit; 3) trafficked to the cell membrane; and 4) incorporated into a functional channel.
Although the N-terminal extracellular domain is important for the initial association between subunits, a downstream domain, the cytoplasmic loop between M1 and M2, participates in the subsequent interaction and stabilization of the oligomeric complex (48). Further downstream, the M3-M4 intracellular loop also affects nAChR assembly (49,50). Assembly of dup␣7 and dup⌬␣7 with full-length ␣7 suggests that other regions of sequence may mediate the interaction between subunits and thus compensate for the lack of the N terminus.
The duplicated subunits also interact with other nAChR subtypes, such as ␣4 and ␣3, suggesting that they may have broad effects on nAChR function, depending on the cell types and brain regions in which they are expressed. Because the N-terminal extracellular domain controls specificity of subunit interaction, future studies should examine whether duplicated subunits assemble with other Cys loop receptors, such as serotonin 5-HT 3 , that do not normally interact with full-length ␣7. We do not know whether the interactions of the duplicated ␣7 subunit, found here in the admittedly forced heterologous system, also occur in human neurons.
Immunostaining of dup␣7 overexpressed in SH-EP cells confirmed that translated peptide contains the sequence for the ␣7 subunit (19). The translation of CHRFAM7A⌬2bp, however, is based only on prediction. When the 2-bp deletion in exon 6 is present, translation is likely to start at one of the two initiating methionines in exon 6. The peptide is out of frame for either 6 or 13 amino acids, depending on which ATG is used, until the 2-bp deletion is reached. At that point, the amino acid sequence returns to the reading frame of the ␣7 subunit. The resulting protein, dup⌬␣7, is smaller than dup␣7 and lacks a larger portion of the agonist-binding domain. We fused a fluorescent protein into the M3-M4 loop of dup⌬␣7 and expressed the labeled protein in Neuro2a cells. Fluorescent signals were observed, and they overlap with the signals from fluorescently labeled full-length ␣7. Our confocal imaging data confirmed that dup⌬␣7 is indeed translated, and the protein is in-frame with ␣7.
We have studied the effect of the duplicated ␣7 genes on ␣7 nAChR function. In theory, they could regulate receptor expression and function in other ways, such as at the RNA level. Furthermore, several ATGs in CHRFAM7A could result in truncated transcripts that do not contain CHRNA7 coding sequence. Because of the frameshift in CHRFAM7A⌬2bp, use of the dup␣7 subunit initiating methionine would encode a unique 40-amino acid peptide, terminated by a stop codon. With no homology to ␣7, it is unlikely that such a truncated peptide would assemble with ␣7 subunits.
The duplicated subunits may have a role in inflammatory responses. Recent studies described ␣7 nAChR as a link in the cholinergic anti-inflammatory pathway and an anti-inflammatory target (51)(52)(53). dup␣7 is highly expressed in macrophages, and it is down-regulated at the mRNA level by IL-1, LPS, and nicotine (20,54). Duplicated subunits could therefore modulate ␣7 receptor-mediated cholinergic anti-inflammatory responses. We noted that co-expression of ␣7 increased the proportion of dup␣7-expressing cells by 2-fold (data not shown); we do not know whether this occurred because ␣7 enhances dup␣7 mRNA expression or acts as a protein chaperone for dup␣7 survival. Thus, investigating dup␣7/␣7 pathways may have medical applications in the treatment of inflammatory disorders.
We found that co-expression of either duplicated subunit with ␣7 subunits in mammalian cells had no marked effects on ACh-induced currents. This finding contrasts with previous studies in oocytes, showing a dominant negative effect of dup␣7 on the full-length receptor (19,20). Many previous studies show that ion channels and receptors assemble differently in these two expression systems, probably in several ways. First, dup␣7 and dup⌬␣7 genes and proteins are expressed at extremely low levels in mammalian cells. Although we used equimolar amounts of ␣7 and dup␣7 DNA for transfection, the actual mRNA and protein ratio of dup␣7/␣7 was probably 1:10 or lower. Because small amounts of dup␣7 subunits were present, few ␣7/dup␣7 heteromers included more than one dup␣7 subunit. A previous study shows that nAChRs containing only a single non-␣7 subunit, and therefore lacking two ␣7-␣7 interfaces, do function rather well (55); but an additional non-␣7 subunit does abolish function (55). This surfeit of dup␣7 or ␣7/dup␣7 and ␣7/dup⌬␣7 Heteropentameric nAChRs dup⌬␣7 subunits may be possible in oocytes, producing the apparent dominant negative effect, but apparently not in the present mammalian system. Second, our studies in Neuro2a cells also included RIC-3 (resistant to inhibitor of cholinesterase) expression. RIC-3 is an endoplasmic reticulum chaperone for nAChRs (56 -58). The effects of the RIC-3 level on ␣7 receptors display an inverted U-shape dose-response relation (57). RIC-3 also shows differential effects when co-expressed with various ligand-gated ion channels as follows: enhancing functional expression of multiple nAChR subtypes and inhibiting 5-HT 3 receptors and several nAChR subtypes, including ␣3␤4, but had no effect on either GABA or glutamate receptors (59 -61). We co-expressed RIC-3 in Neuro2a cells for all our electrophysiological experiments to enable functional expression of ␣7 on the cell membrane. However, we do not know what influence RIC-3 may have on duplicated subunits and how RIC-3 may regulate the effect of duplicated subunits on full-length ␣7 receptors.
Although we did not observe a significant difference in the ACh response of ␣7 receptors when dup␣7 or dup⌬␣7 was co-expressed, dup⌬␣7/␣7 did display a lower choline sensitivity than dup␣7/␣7 heteromers. In addition, dup␣7/␣7 or dup⌬␣7/␣7 receptors desensitize more quickly during brief (1 to 2 s) exposure to physiological concentrations of choline. Genetic and neurobiological studies suggest that choline availability may contribute to the development of schizophreniaassociated sensory gating deficits (62), and perinatal dietary choline supplementation has been tested for lowering schizophrenia risk, with positive results (63). The copy of the duplicated gene with the 2-bp deletion, CHRFAM7A⌬2bp, is found much more frequently in schizophrenic patients than in control subjects with no mental illness (16) and is associated with abnormal sensory processing (64). Our electrophysiological data, suggesting that dup⌬␣7 lacks the augmentation of choline sensitivity produced by dup␣7, may be relevant to a decreased efficacy of choline during development in subjects that carry this 2-bp deletion.